Fix #6559: Wrong node-ids are generated in glossary directive

5 files changed, 53 insertions, 24 deletions

diff --git a/CHANGES b/CHANGES
index 946870089..fa0b2a2b1 100644
--- a/CHANGES
+++ b/CHANGES
@@ -12,6 +12,7 @@ Deprecated
 
 * The ``decode`` argument of ``sphinx.pycode.ModuleAnalyzer()``
 * ``sphinx.directives.other.Index``
+* ``sphinx.environment.temp_data['gloss_entries']``
 * ``sphinx.environment.BuildEnvironment.indexentries``
 * ``sphinx.environment.collectors.indexentries.IndexEntriesCollector``
 * ``sphinx.io.FiletypeNotFoundError``
@@ -36,6 +37,7 @@ Bugs fixed
 * #6925: html: Remove redundant type="text/javascript" from <script> elements
 * #6906, #6907: autodoc: failed to read the source codes encoeded in cp1251
 * #6961: latex: warning for babel shown twice
+* #6559: Wrong node-ids are generated in glossary directive
 
 Testing
 --------
diff --git a/doc/extdev/deprecated.rst b/doc/extdev/deprecated.rst
index f1c03ec54..36b33e829 100644
--- a/doc/extdev/deprecated.rst
+++ b/doc/extdev/deprecated.rst
@@ -36,6 +36,11 @@ The following is a list of deprecated interfaces.
      - 4.0
      - ``sphinx.domains.index.IndexDirective``
 
+   * - ``sphinx.environment.temp_data['gloss_entries']``
+     - 2.4
+     - 4.0
+     - ``documents.nameids``
+
    * - ``sphinx.environment.BuildEnvironment.indexentries``
      - 2.4
      - 4.0
diff --git a/sphinx/domains/std.py b/sphinx/domains/std.py
index 909714a4a..2d2b47499 100644
--- a/sphinx/domains/std.py
+++ b/sphinx/domains/std.py
@@ -243,34 +243,50 @@ def split_term_classifiers(line: str) -> List[Optional[str]]:
 
 
 def make_glossary_term(env: "BuildEnvironment", textnodes: Iterable[Node], index_key: str,
-                       source: str, lineno: int, new_id: str = None) -> nodes.term:
+                       source: str, lineno: int, node_id: str = None,
+                       document: nodes.document = None) -> nodes.term:
     # get a text-only representation of the term and register it
     # as a cross-reference target
     term = nodes.term('', '', *textnodes)
     term.source = source
     term.line = lineno
-
-    gloss_entries = env.temp_data.setdefault('gloss_entries', set())
     termtext = term.astext()
-    if new_id is None:
-        new_id = nodes.make_id('term-' + termtext)
-        if new_id == 'term':
-            # the term is not good for node_id.  Generate it by sequence number instead.
-            new_id = 'term-%d' % env.new_serialno('glossary')
-    while new_id in gloss_entries:
-        new_id = 'term-%d' % env.new_serialno('glossary')
-    gloss_entries.add(new_id)
+
+    if node_id:
+        # node_id is given from outside (mainly i18n module), use it forcedly
+        pass
+    elif document:
+        node_id = nodes.make_id('term-' + termtext)
+        if node_id == 'term':
+            # "term" is not good for node_id.  Generate it by sequence number instead.
+            node_id = 'term-%d' % env.new_serialno('glossary')
+
+        while node_id in document.ids:
+            node_id = 'term-%d' % env.new_serialno('glossary')
+
+        document.note_explicit_target(term)
+    else:
+        warnings.warn('make_glossary_term() expects document is passed as an argument.',
+                      RemovedInSphinx40Warning)
+        gloss_entries = env.temp_data.setdefault('gloss_entries', set())
+        node_id = nodes.make_id('term-' + termtext)
+        if node_id == 'term':
+            # "term" is not good for node_id.  Generate it by sequence number instead.
+            node_id = 'term-%d' % env.new_serialno('glossary')
+
+        while node_id in gloss_entries:
+            node_id = 'term-%d' % env.new_serialno('glossary')
+        gloss_entries.add(node_id)
+        term['ids'].append(node_id)
 
     std = cast(StandardDomain, env.get_domain('std'))
-    std.add_object('term', termtext.lower(), env.docname, new_id)
+    std.add_object('term', termtext.lower(), env.docname, node_id)
 
     # add an index entry too
     indexnode = addnodes.index()
-    indexnode['entries'] = [('single', termtext, new_id, 'main', index_key)]
+    indexnode['entries'] = [('single', termtext, node_id, 'main', index_key)]
     indexnode.source, indexnode.line = term.source, term.line
     term.append(indexnode)
-    term['ids'].append(new_id)
-    term['names'].append(new_id)
 
     return term
 
@@ -368,7 +384,8 @@ class Glossary(SphinxDirective):
                 textnodes, sysmsg = self.state.inline_text(parts[0], lineno)
 
                 # use first classifier as a index key
-                term = make_glossary_term(self.env, textnodes, parts[1], source, lineno)
+                term = make_glossary_term(self.env, textnodes, parts[1], source, lineno,
+                                          document=self.state.document)
                 term.rawsource = line
                 system_messages.extend(sysmsg)
                 termtexts.append(term.astext())
diff --git a/sphinx/transforms/i18n.py b/sphinx/transforms/i18n.py
index e61905830..34d5b1368 100644
--- a/sphinx/transforms/i18n.py
+++ b/sphinx/transforms/i18n.py
@@ -202,18 +202,13 @@ class Locale(SphinxTransform):
 
             # glossary terms update refid
             if isinstance(node, nodes.term):
-                gloss_entries = self.env.temp_data.setdefault('gloss_entries', set())
-                for _id in node['names']:
-                    if _id in gloss_entries:
-                        gloss_entries.remove(_id)
-
+                for _id in node['ids']:
                     parts = split_term_classifiers(msgstr)
                     patch = publish_msgstr(self.app, parts[0], source,
                                            node.line, self.config, settings)
                     patch = make_glossary_term(self.env, patch, parts[1],
-                                               source, node.line, _id)
-                    node['ids'] = patch['ids']
-                    node['names'] = patch['names']
+                                               source, node.line, _id,
+                                               self.document)
                     processed = True
 
             # update leaves with processed nodes
diff --git a/tests/test_domain_std.py b/tests/test_domain_std.py
index 538bd9cde..56c2d968c 100644
--- a/tests/test_domain_std.py
+++ b/tests/test_domain_std.py
@@ -255,6 +255,16 @@ def test_glossary_alphanumeric(app):
     assert ("/", "/", "term", "index", "term-0", -1) in objects
 
 
+def test_glossary_conflicted_labels(app):
+    text = (".. _term-foo:\n"
+            ".. glossary::\n"
+            "\n"
+            "   foo\n")
+    restructuredtext.parse(app, text)
+    objects = list(app.env.get_domain("std").get_objects())
+    assert ("foo", "foo", "term", "index", "term-0", -1) in objects
+
+
 def test_cmdoption(app):
     text = (".. program:: ls\n"
             "\n"